[PATCH] Don't shutdown mpc5200 serial port if it is a console

[PATCH] Don't shutdown mpc5200 serial port if it is a console

[Grant Likely]

If the serial port gets shut down, then console output stalls.  9 out
of 10 kernel hackers agree, this is a bad thing.

Signed-off-by: Grant Likely <grant.likely@secretlab.ca>
Signed-off-by: Bartlomiej Sieka <tur@semihalf.com>
---

This is a simple bug fix, but it's not a critical one.  It would be really
nice to get this merged for 2.6.21
 drivers/serial/mpc52xx_uart.c |   14 ++++++++------
 1 files changed, 8 insertions(+), 6 deletions(-)

diff --git a/drivers/serial/mpc52xx_uart.c b/drivers/serial/mpc52xx_uart.c
index f8c1761..35669f8 100644
--- a/drivers/serial/mpc52xx_uart.c
+++ b/drivers/serial/mpc52xx_uart.c
@@ -257,14 +257,16 @@ mpc52xx_uart_shutdown(struct uart_port *port)
 {
 	struct mpc52xx_psc __iomem *psc = PSC(port);
 
-	/* Shut down the port, interrupt and all */
-	out_8(&psc->command,MPC52xx_PSC_RST_RX);
-	out_8(&psc->command,MPC52xx_PSC_RST_TX);
+	/* We don't want to shut down a port that is used as a console */
+	if (!uart_console(port)) {
+		out_8(&psc->command,MPC52xx_PSC_RST_RX);
+		out_8(&psc->command,MPC52xx_PSC_RST_TX);
 
-	port->read_status_mask = 0;
-	out_be16(&psc->mpc52xx_psc_imr,port->read_status_mask);
+		port->read_status_mask = 0;
+		out_be16(&psc->mpc52xx_psc_imr,port->read_status_mask);
+	}
 
-	/* Release interrupt */
+	/* Release interrupt (OK for console port b/c console uses polling) */
 	free_irq(port->irq, port);
 }
 
-- 
1.5.1

Re: [PATCH] Don't shutdown mpc5200 serial port if it is a console

[Sylvain Munaut]

The idea is OK,

but only TX should be left active and the interrupts should still be
disabled (imr set to 0), so we don't generate unhandled interrupts ...


        Sylvain


Grant Likely wrote:
> If the serial port gets shut down, then console output stalls.  9 out
> of 10 kernel hackers agree, this is a bad thing.
>
> Signed-off-by: Grant Likely <grant.likely@secretlab.ca>
> Signed-off-by: Bartlomiej Sieka <tur@semihalf.com>
> ---
>
> This is a simple bug fix, but it's not a critical one.  It would be really
> nice to get this merged for 2.6.21
>  drivers/serial/mpc52xx_uart.c |   14 ++++++++------
>  1 files changed, 8 insertions(+), 6 deletions(-)
>
> diff --git a/drivers/serial/mpc52xx_uart.c b/drivers/serial/mpc52xx_uart.c
> index f8c1761..35669f8 100644
> --- a/drivers/serial/mpc52xx_uart.c
> +++ b/drivers/serial/mpc52xx_uart.c
> @@ -257,14 +257,16 @@ mpc52xx_uart_shutdown(struct uart_port *port)
>  {
>  	struct mpc52xx_psc __iomem *psc = PSC(port);
>  
> -	/* Shut down the port, interrupt and all */
> -	out_8(&psc->command,MPC52xx_PSC_RST_RX);
> -	out_8(&psc->command,MPC52xx_PSC_RST_TX);
> +	/* We don't want to shut down a port that is used as a console */
> +	if (!uart_console(port)) {
> +		out_8(&psc->command,MPC52xx_PSC_RST_RX);
> +		out_8(&psc->command,MPC52xx_PSC_RST_TX);
>  
> -	port->read_status_mask = 0;
> -	out_be16(&psc->mpc52xx_psc_imr,port->read_status_mask);
> +		port->read_status_mask = 0;
> +		out_be16(&psc->mpc52xx_psc_imr,port->read_status_mask);
> +	}
>  
> -	/* Release interrupt */
> +	/* Release interrupt (OK for console port b/c console uses polling) */
>  	free_irq(port->irq, port);
>  }
>  
>